How artificial intelligence is changing the world of defibrillators
Success rates have improved, but the cost and high computing power are a problem.
Defibrillators are used to deliver electrical current to the heart as a treatment for potentially fatal cardiac arrest. Artificial intelligence is having a huge impact on how defibrillators work more effectively, with machine learning algorithms becoming more accurate in delivering life-saving treatments, according to a new paper.

Automated external defibrillators (AEDs) and implantable cardioverter defibrillators (ICDs) use shock warning algorithms to identify echocardiographic recordings. The data determines whether rhythms are considered “shockable” or “shocks” to determine if defibrillation was necessary for treatment.

Artificial intelligence can also be used to diagnose the causes of heart attacks, classify heart rhythms that do not interfere with cardiopulmonary resuscitation (CPR), and predict the success of defibrillation, according to “Role of Artificial Intelligence in Defibrillators: A Narrative Review” by researchers at a UK hospital and university. As success rates have improved, cost and high computational power remain an issue.

The way machine learning is implemented in medical applications has evolved in recent years. Currently, supervised machine learning models are still needed for defibrillator applications. Deep learning mimics the brain’s neural networks using artificial neural networks (ANNs) with layers of nodes that process input data.

AI can be used for standard ECG analysis. Convolutional neural networks (CNNs) are a subcategory of ANNs that use high-level features of raw data. This method is used in medical imaging and ECG analysis, but can also be used to evaluate multiple dimensions of data sets. In one model, CNN’s ECG interpretation is more accurate than that of human cardiologists, but automated ECG analysis is still not widely used.

Smartwatches such as the Apple Watch and other wearable technologies are increasingly used, particularly with the ability to perform automated single-lead ECGs to detect atrial fibrillation.

In one study, the Kardia Band (KB) algorithm used by the Apple Watch was not as accurate in diagnosing clients. The algorithm was unable to interpret more than half of the ECG and CB recordings. A doctor’s supervision is still necessary for the most accurate diagnosis.

In addition, AI can be used as a screening for early pulmonary hypertension and asymptomatic left ventricular dysfunction. Mayo Clinic used data from nearly 45,000 patients to train CNN to identify asymptomatic left ventricular dysfunction. The results are promising and a positive AI screening predicts a four times higher risk of developing ventricular dysfunction than in patients who did not use the screen. Artificial intelligence can also control when to stop CPR and optimize shock delivery. It can also reduce the number of unnecessary shocks, which can improve patient viability. Future plans include creating a robust ECG dataset to build and test technology comparison algorithms.